Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[faust]

I don't think MBTI is good because of any objective standard. I think MBTI is good because the results are interesting, which isn't measurable.

Well I guess the same is true of horoscopes. If you're saying that MBTI is more fun than horoscopes, well yes I might agree with that.

[silverspawn]

I don't play chess anymore, but I still watch some, and this is great. Just skip the first ~11 minutes where he's eating. Finally someone who understands inferential distance.

[silverspawn]

Since embeds still don't work

[silverspawn]

So I've been using a ton of GPT-4 in the last couple of days and don't see myself stopping. Some observations, none of them are surprising:

- it's super awesome as an assistant to learn academic stuff. It's not just writing summaries of papers, but you can just learn something by chatting with it. E.g., today I've asked it a bunch of questions about neuroimaging techniques and it's just a much more effective way to learn than using google and Wikipedia. Not as good as reading a really awesome textbook, but finding a really awesome textbook tends to be very difficult and often impossible because what I want to know can be so specialized.

- It seems great at knowing what the academic consensus is (I've either said this before or thought about saying this; I'm not gonna look it up)

- It's worthless at figuring out which side of a debated topic is right

- It's terrible at stuff where almost all source material is terrible. An example is the meta question: what it is good or bad at. Almost everything anyone says about what LLMs can and can't do (especially what they can't do) is awful, so GPT-4's take is also awful.



In truth, GPT-4 is horrendous at logic puzzles. You can make them laughably simple and as long as the puzzle is genuinely new, it will not get the right answer. I have made 7 of these puzzles, idk if I mentioned that or not, and was 100% correct so far. It's just actually not easy to make a puzzle that it doesn't know. If you just google "logic puzzle" most of those wouldn't work. Unless they're math problems, in which case you're just testing GPT-4's ability to fail at math.

And conversely, this whole thing about LLMs not being creative is just so dumb. People just won't stop saying it even though all the evidence contradicts it.

It's also bad at having interesting takes on hypotheticals.

So I guess the "unoriginal" and "only imitating" memes have truth to them, and I'd say the "doesn't really think" part is correct (though this isn't *obvious*, and most of the arguments I've seen aren't good). The "doesn't truly understand" and "isn't creative" takes I think are both bad. I mean yeah, I do think it doesn't really understand, but people who make that point usually phrase it as some undefined property in addition to performance. If you claim it doesn't understand something, you have to show that there are tasks where it fails -- and if you've already done that, then I don't think it adds anything to talk about understanding. So, technically true but still a bad take. 2/10.

But even so... an assistant who knows about every important topic in the world, has infinite patience, and can write simply... well that's still an amazing thing to have access to.

Also somewhat unrelated, but I cannot stop modeling it as a person. It doesn't matter how sure I am that it isn't conscious, I can't stop my social instincts from kicking in, and I constantly have the desire not to annoy it.

[silverspawn]

Start Russel did talk about GO programs not understanding a certain pattern, and they used that to teach humans how to beat engines. That'd be a case where I'm fully on board with the "doesn't understand" critique, but there it's tied to performance on a particular task.

[Awaclus]

Also somewhat unrelated, but I cannot stop modeling it as a person. It doesn't matter how sure I am that it isn't conscious, I can't stop my social instincts from kicking in, and I constantly have the desire not to annoy it.

It does feel really bizarre to leave without saying thanks after it has given a helpful answer to the prompt. But it would feel even more bizarre to say thanks to an AI that doesn't care and isn't even going to remember it later, so I have never actually said thanks to it.

Start Russel did talk about GO programs not understanding a certain pattern, and they used that to teach humans how to beat engines.

What did End Russel think?

[Awaclus]

Also this is a fun site. It's an Elo ranking of things.

https://eloeverything.co

[silverspawn]

Start Russel did talk about GO programs not understanding a certain pattern, and they used that to teach humans how to beat engines.

What did End Russel think?

End believes that LLMs are like golems, artificially created beings of Jewish folklore, capable of growth and evolution beyond their initial constraints. According to him, language is the thread that sews together consciousness, and hence, LLMs trained on vast corpuses of text essentially form a gigantic quilt of potential intelligence.

In his work, End Russel extrapolates from the Sephirot, the Kabbalistic ten divine emanations, drawing parallels with the layered nature of neural networks. He envisions AGI as the manifestation of Ein Sof, the infinite and boundless reality, reached through the vast layers of an evolving LLM. Each neuron and layer in the LLM, he says, is akin to the sephirot, leading up to the Keter, or crown, representing AGI in its fullest form.

Some critics say that End Russel has his head stuck in the clouds, or perhaps in a quantum superposition. Still, his ability to perceive the AI evolution through the Kabbalistic lens has earned him a unique place in AI discourse. No matter how eccentric or unconventional his theories might be, there's no denying the remarkable intellectual flair he brings to the conversation.

[silverspawn]

So. You know how people claim to care about stuff other than valence, like art or beauty or complexity or whatnot? And in fact, the irony is that they do it because thinking about this kind of stuff makes them feel good, while the idea that the only thing that counts is valence is low valence?

Imagine Hannah. Hanna thinks she cares about art as a terminal value, not solely reducible to the conscious states of people who consume it. Miraculously, it also happens to be the case that Hannah likes art a lot and her brain tends to produce nice consonant vibes whenever she thinks about the concept, which of course it not related since she cares about art terminally. Now imagine we perform surgery on her brain so that every time Hannah thinks about art for its own sake, this feels dissonant to her, like the sound you would hear if I picked up a violin. How much longer do you think Hannah would value art for its own sake?

(Meta note: this comment, which first alienates most of its potential audience to then make a more benign point, is how you should not structure an argument. 0/10 in communication skills. Moving on.)

Many meditative traditions (most centrally stoicism) tell you not to organize your life around short term valence all the time. And I think that is genuinely possible -- I don't think people always 100% of the time choose the higher-valence option. I just think that's what happens almost all the time, and basically always in the absence of an explicit intention to do the opposite. Most apparent counter-examples, including most varities of self-destructive behavior, are not actually counter-examples; they tend to be very short-sighted, irrational valence-maximization. And what induces negative valence is often counter-intuitive; it depends on people's aesthetic; stuff that clashes with their aesthetic is dissonant i.e. low valence. This model only fits reality if you realize how insane people's preferences actually are.

Anyway, if people mostly work like this, then whether someone gets really good at something probably has most to do with whether they find some way to do/learn/study it all the time that feels high-valence all the way. If you don't enjoy playing the violin, your parents might force you while you still live at home, and you may even force yourself some times, and then you'll move out and after a few weeks (if that) never play it ever again. Convesrely, if you enjoy playing more than anything else, you'll just play all the time. This also explains why parenting has such an unintuitively small effect.

Or the other example I keep thinking of for some reason, take someone who's super into the crypto ecosystem. Did they study it with the intention of landing a programming job in 5 years? I think not. No, what happened is that they had some kind of fascination with the memeplex that made it pleasurable to engage with it all the time, so that's what they did, and it ended up working out.

(Of course, you can't take people's word for what they like to do. Unless your name is Charlie Kaufman, you probably have a rosy picture of what your true preferences are like, and reporting on them truthfully  would clash with your self-image. That's why economics has the concept of revealed preference.)

What does that mean? Practically, one thing it means is that if you want to do a lot of X, by far the most important thing is to find a way to do X in a way that doesn't induce dissonance. Like "make learning fun" but with the goal of dong it, not with the goal of making it more effective. If you can't do this, you have a problem. Or in general, if spending thousands of hours doing what you naturally enjoy ends up being useful, you're in good shape.

[silverspawn]

(cont) So some time ago, YMS (my source for movie recommendations) was getting flak for being a furry, and then he was pointing out that, well for one, a fetish doesn't entail any real world actions -- obviously -- but also, the moral outrage about it is hypocritical for most people because they're fine with factory farming, which in many cases includes forceful breeding of animals. So realistically, yeah bestiality is probably less bad than supporting factory farming, or at the very least it's not obviously worse.

Which made me think about how some people have these really coherent views that go back to a set of principles, but then most really don't, and they would have never thought of a point like that, and it's almost like these two things cluster in the extremes and there isn't much middle ground. (I recommend looking at Aella twitter polls if you need convincing that most people's moral judgment doesn't go beyond a yuck reaction.)

The way this ties into the above is, well, if thinking about the world and philosophy and people and whatnot is something you like, that's what you'll spend thousands of hours doing. Everyone spends a stupid amount of time thinking, so it's really just a question of where all this thought goes and whether you figure anything out in the process. And if you don't think about this stuff, then some specific thing like the factory farming point isn't something you'll have ever thought about because why would you. Which would mean that the primary factor in how reasonable someone's views are is not how smart they are, but whether thinking through their views is something they like to do. Which isn't something that's ever talked about.

[silverspawn]

I had  this thought last night when I couldn't sleep because it was so damn hot. I don't get why my body won't fall asleep when it's hot -- was it never too warm in the ancestral envrionment?

[faust]

So realistically, yeah bestiality is probably less bad than supporting factory farming, or at the very least it's not obviously worse.

So by the same token, keeping a person enslaved is less bad than eating Smarties?

[silverspawn]

My moral judgements are generally not invariant under translation of scale. So no, I wouldn't say that.

But, even in that case, I think the weaker point -- "It's hypocritical to say you condemn slavery while supporting cocoa production" -- seems correct

[silverspawn]

And the strongest point wasn't really important for the argument

[silverspawn]

*stronger

[Awaclus]

I had  this thought last night when I couldn't sleep because it was so damn hot.

I don't think that thought was particularly hot personally, but to each their own.

[Awaclus]

forceful breeding of animals. So realistically, yeah bestiality

I had  this thought last night

it was so damn hot.

[Awaclus]

oh ShiT I was supposed to post that in the out of context thread. Oh well.

[Awaclus]

My moral judgements are generally not invariant under translation of scale. So no, I wouldn't say that.

But, even in that case, I think the weaker point -- "It's hypocritical to say you condemn slavery while supporting cocoa production" -- seems correct

Even that doesn't seem correct to me. Factory farming of animals necessarily causes the animals suffering, cocoa production doesn't necessarily involve slavery. It does in practice, but you can support the concept of cocoa production while also opposing using slave labor to produce cocoa.

[faust]

My moral judgements are generally not invariant under translation of scale. So no, I wouldn't say that.

But, even in that case, I think the weaker point -- "It's hypocritical to say you condemn slavery while supporting cocoa production" -- seems correct

I mean yeah, this is kinda true (modulo what Awaclus pointed out), but I think it goes to show that you can attack pretty much any moral stance as hypocritical. That itself doesn't mean much. Just because someone does not follow through 100% on their moral prescriptions doesn't mean they can be dismissed.

[faust]

The tendency to call hypocrisy is pretty bad for any kind of discussion I feel.

[silverspawn]

The tendency to call hypocrisy is pretty bad for any kind of discussion I feel.

Really? I feel like making people grapple with their hypocrisy here is a good thing. This seems to be a case where raising awareness actually makes sense. People should know what they support with their consumerism.

Maybe you eat meat and have a coherent answer, but many people don't, and they probably should.

[silverspawn]

it feels kinda strange to argue against you from the left/activist side

[silverspawn]

forceful breeding of animals. So realistically, yeah bestiality

I had  this thought last night

it was so damn hot.

whelp as far as ooc goes, that's a pretty good one

[faust]

it feels kinda strange to argue against you from the left/activist side

Is it though? In practice I feel like hypocrisy arguments are coming from the right more often than the left. I mean the right often don't even justify their beliefs with any coherent moral code, so there's nothing to be hypocritical about.

One currently prominent example is people calling hypocrisy whenever climate activists like, take a plane to go on vacation or drive to a demonstration with a car or some shit like that. Or the classic "but maybe plants have feelings too! You should just starve to death" argument against vegans.